Semiconductor memory devices are becoming more and more complex as their size decreases and their storage density increases. To help handle some of the increase in storage density, an architecture comprising multiple subarrays of memory cells on a die for storing values such as bits has been adopted in dynamic random access memory (DRAM) devices. Each of the subarrays comprises multiple rows of memory cells that are accessed or "fired" by activation of row address signals. Occasionally, during manufacture, one or more rows is defective. Some of these rows may be replaced via a fuse option with redundant rows such as shown in U.S. Pat. No. 5,528,539 to Ong et. al. When a redundant row is used, the DRAM's internal timing needs to be slightly slowed down, to provide extra time for address compare and override to the redundant rows. Repair of DRAMs during manufacturing happens on a manageably small percentage of parts, often less than 50%. Therefore, it is not desirable to slow down every die regardless of row repair. In fact, it is desired to obtain faster row access speeds if possible.
Prior solutions have included providing circuits on the die that poll every redundant row bank on the die. If any are enabled, the RAS chain is slowed down. Such schemes, while easy to implement on smaller generation DRAMs, mandate a large number of line spaces and gates for the polling operation on higher density generation DRAMS, since there may be 16 to 64 row banks or more that are checked. This consumes valuable die space and adversely impacts efforts to further increase DRAM density.
There is a need for slowing down the RAS chain when redundant rows are used. There is a further need to only slow down the RAS chain when such redundant rows are in use. There is a need to slow down the RAS chain without using significant additional circuitry. In fact, it would be beneficial to reduce the amount of circuitry currently used to determine that the RAS chain needs to be slowed down.